PCA using Matlab, SPSS

Hi

Following is the program I run for PCA. For the same set of data I tried PCA with various rotations in SPSS. But nothing gave the same plot as matlab did. Am I making any mistake in the program or why is the difference?

A=xlsread('sn30.xlsx');A = A';
[n m] = size(A);
AMean = mean(A);
AStd = std(A);
B = (A - repmat(AMean,[n 1])) ./ repmat(AStd,[n 1]);
[COEFF SCORE LATENT TSquare] = princomp(B);
 PC1=SCORE;
A1=PC1(:,1);
 B1=PC1(:,2);
 C1=PC1(:,3);
 for ii=1:8
     plot3(A1(ii,1),B1(ii,1),C1(ii,1),'r*'); hold on;
 end
 for ii=9:15

       plot3(A(ii,1),B1(ii,1),C1(ii,1),'bo'); hold on;
  end
   for ii=16:25

       plot3(A(ii,1),B1(ii,1),C1(ii,1),'g^'); hold on;
  end

Regards

Amutha

Answer 1

[Aditya]

Hi Amrutha,

When comparing PCA results between MATLAB and SPSS, there are several factors that can lead to differences in the plots, even when using the same dataset.Here's a cleaned-up version of your MATLAB code using pca, which is more modern and recommended over princomp:

% Read and preprocess data
A = xlsread('sn30.xlsx'); 
A = A';
[n, m] = size(A);

% Standardize the data
AMean = mean(A);
AStd = std(A);
B = (A - AMean) ./ AStd;

% Perform PCA
[coeff, score, latent, tsquare] = pca(B);

% Extract principal component scores
PC1 = score(:, 1);
PC2 = score(:, 2);
PC3 = score(:, 3);

% Plot the first three principal components
figure;
hold on;
for ii = 1:8
    plot3(PC1(ii), PC2(ii), PC3(ii), 'r*');
end
for ii = 9:15
    plot3(PC1(ii), PC2(ii), PC3(ii), 'bo');
end
for ii = 16:25
    plot3(PC1(ii), PC2(ii), PC3(ii), 'g^');
end
xlabel('PC1');
ylabel('PC2');
zlabel('PC3');
title('PCA Plot');
grid on;
hold off;